如何访问字典中的键？

Question

作为 python 的初学者，我正在处理一个 CSV 文件并创建了一个边缘列表，以便这些值与 CSV 中该行的值具有一对一的映射，例如： 输出：

value1 value2     
value1 value3
value2 value3    
value4 value5
.
.
.

然后我为 csv 文件中的每个值分配了唯一编号，以便唯一编号充当键，而 CSV 中的项目充当字典中的值。此外，如果任何值在 CSV 文件中重复，我不想为其分配另一个键。

输出：

dictionary=
{
1: "value1",
2: "value2",
3: "value3",
.
.
.
}

现在我想要边缘列表（我之前创建的）作为输出，但是边缘列表中的值应该被它们在字典中的键替换，例如：

1 2
1 3
2 3
.
.
.

谢谢！

Answer 1

我想这里的主要问题是在构建从唯一数字到名称的字典时，因为我认为您应该以相反的方式构建它（从名称到唯一数字）。此外，在构建此地图时，您缺少将映射转换为具有唯一编号的最终映射的代码。

在下面找到我建议的代码：

#!/usr/bin/python                                                                                                   

# -*- coding: utf-8 -*-

# For better print formatting
from __future__ import print_function

# Imports
import sys


#
# HELPER METHODS
#

def mapping(csv_filename, mapping_filename):
    if __debug__:
        print("CSV File: " + str(csv_filename))
        print("Mapping File: " + str(mapping_filename))

    # Retrieve data from CSV file
    with open(csv_filename, "r") as csv_file:
        data_raw = csv_file.readlines()
    data = []
    for line in data_raw:
        line = line.strip()
        elements = line.split(",")
        elements = [e.strip() for e in elements]
        data.append(elements)

    # Create mapping list and file
    mapping_list = []
    with open(mapping_filename, "w") as mapping_file:
        for elements in data:
            j = 0
            while j != len(elements) - 1:
                for k in range(j + 1, len(elements)):
                    # Add to mapping
                    temp = [elements[j], elements[k]]
                    mapping_list.append(temp)
                    # Write to file
                    mapping_file.write(elements[j] + " " + elements[k] + "\n")
                j += 1

    # Return the mapping
    return mapping_list


def build_key_map(mapping_list):
    if __debug__:
        print("Mapping List: " + str(mapping_list))

    key_dict = {}
    i = 1
    # Check each parsed node inside each edge
    for edge in mapping_list:
        for node in edge:
            # Add node to keys if it has not been registered yet
            if node not in key_dict.keys():
                key_dict[node] = i
                i = i + 1

    return key_dict


def build_graph(mapping_list, key_dict):
    if __debug__:
        print("Mapping List: " + str(mapping_list))
        print("Key Dict: " + str(key_dict))

    # Copy the existing mapping changing each node (inside edge) by its unique number
    new_mapping_list = []
    for edge in mapping_list:
        new_edge = []
        for node in edge:
            new_edge.append(key_dict[node])
        new_mapping_list.append(new_edge)

    return new_mapping_list


#
# MAIN
#

def main():
    import sys
    csv_file = sys.argv[1]
    mapping_file = sys.argv[2]

    mapping_list = mapping(csv_file, mapping_file)
    key_dict = build_key_map(mapping_list)
    new_mapping_list = build_graph(mapping_list, key_dict)

    print("FINAL MAPPING: ")
    for edge in new_mapping_list:
        print(edge)


#
# ENTRY POINT
#

if __name__ == "__main__":
    main()

请注意，我已尝试将您的代码保存在单独的函数中，尽管将所有进程放在一个函数中可能会提高性能：

• mapping function 解析 CSV 文件并生成映射（从节点到节点的边列表）并将其写入给定文件。在这里，CSV 解析由“，”（如您提供的示例）分隔，尽管您的原始代码将其分隔为“：”。
• build_key_map 创建一个从节点名称到唯一编号的字典。
• build_graph 使用节点名称的唯一编号转换映射。

使用您的输入，预期的输出是：

> python parser.py "csv.txt" "map.txt"
CSV File: csv.txt
Mapping File: map.txt
Mapping List: [['man', 'nut'], ['man', 'bag'], ['nut', 'bag'], ['rat', 'cat'], ['dog', 'fog'], ['dog', 'cat'], ['dog', 'man'], ['fog', 'cat'], ['fog', 'man'], ['cat', 'man']]
Mapping List: [['man', 'nut'], ['man', 'bag'], ['nut', 'bag'], ['rat', 'cat'], ['dog', 'fog'], ['dog', 'cat'], ['dog', 'man'], ['fog', 'cat'], ['fog', 'man'], ['cat', 'man']]
Key Dict: {'nut': 2, 'dog': 6, 'cat': 5, 'bag': 3, 'rat': 4, 'fog': 7, 'man': 1}
FINAL MAPPING: 
[1, 2]
[1, 3]
[2, 3]
[4, 5]
[6, 7]
[6, 5]
[6, 1]
[7, 5]
[7, 1]
[5, 1]

另外，我添加了一些 cmets，但如果我需要澄清某些部分，请告诉我。

编辑：

顺便说一句，如果您真的需要从唯一数字到值的映射，您可以随时反转字典并将其存储，而算法可以继续处理从名称到唯一数字的字典。要反转字典，您只需：

inverted_dict = dict([[v,k] for k,v in key_dict.items()])

EDIT2：

我在这里提供了另一个版本的映射函数，它直接生成具有唯一值的映射（而不是使用多个函数和中间结构）。

def mapping(csv_filename, mapping_filename):
    if __debug__:
        print("CSV File: " + str(csv_filename))
        print("Mapping File: " + str(mapping_filename))

    # Retrieve data from CSV file
    with open(csv_filename, "r") as csv_file:
        data_raw = csv_file.readlines()
    data = []
    for line in data_raw:
        line = line.strip()
        elements = line.split(",")
        elements = [e.strip() for e in elements]
        data.append(elements)

    # Create mapping list and file
    mapping_list = []
    key_dict = {}
    unique_num = 1
    with open(mapping_filename, "w") as mapping_file:
        for elements in data:
            j = 0
            while j != len(elements) - 1:
                for k in range(j + 1, len(elements)):
                    if __debug__:
                        print("Converting: " + elements[j] + " -> " + elements[k])
                    # Transform elements to keys
                    if elements[j] in key_dict.keys():
                        key_j = key_dict[elements[j]]
                    else:
                        key_dict[elements[j]] = unique_num
                        key_j = unique_num
                        unique_num = unique_num + 1
                    if elements[k] in key_dict.keys():
                        key_k = key_dict[elements[k]]
                    else:
                        key_dict[elements[k]] = unique_num
                        key_k = unique_num
                        unique_num = unique_num + 1
                    # Add to mapping
                    if __debug__:
                        print("Adding: " + str(key_j) + " -> " + str(key_k))
                    mapping_list.append([key_j, key_k])
                    # Write to file
                    mapping_file.write(str(key_j) + " " + str(key_k) + "\n")
                j += 1

    # Return the mapping
    return mapping_list

它的预期输出是：

> python parser.py "csv.txt" "map.txt"
CSV File: csv.txt
Mapping File: map.txt
Converting: man -> nut
Adding: 1 -> 2
Converting: man -> bag
Adding: 1 -> 3
Converting: nut -> bag
Adding: 2 -> 3
Converting: rat -> cat
Adding: 4 -> 5
Converting: dog -> fog
Adding: 6 -> 7
Converting: dog -> cat
Adding: 6 -> 5
Converting: dog -> man
Adding: 6 -> 1
Converting: fog -> cat
Adding: 7 -> 5
Converting: fog -> man
Adding: 7 -> 1
Converting: cat -> man
Adding: 5 -> 1
FINAL MAPPING: 
[1, 2]
[1, 3]
[2, 3]
[4, 5]
[6, 7]
[6, 5]
[6, 1]
[7, 5]
[7, 1]
[5, 1]

Answer 2

除非别无选择，否则不应在循环中手动管理索引 (while(j!=len(i)-1))。

对于组合，您可以使用itertools：

>>> import itertools
>>> list(itertools.combinations(["man", "nut", "bag"], 2))
[('man', 'nut'), ('man', 'bag'), ('nut', 'bag')]

然后很容易获得边缘。

我为您提供的示例创建了一个阅读器：

>>> data = """man,nut,bag
... rat,cat
... dog,fog,cat,man"""
...
>>> import io
>>> import csv
>>> reader = csv.reader(io.StringIO(data))

边是每一行组合的串联：

>>> edges = [(v1, v2) for row in reader for v1, v2 in itertools.combinations(row, 2)]
>>> edges
[('man', 'nut'), ('man', 'bag'), ('nut', 'bag'), ('rat', 'cat'), ('dog', 'fog'), ('dog', 'cat'), ('dog', 'man'), ('fog', 'cat'), ('fog', 'man'), ('cat', 'man')]

现在您可以从edges 中提取独特的元素：

>>> vs = sorted(set(a for e in edges for a in e))
>>> vs
['bag', 'cat', 'dog', 'fog', 'man', 'nut', 'rat']

（我在这里使用sorted 来获得可重现的结果，但您不需要它）。要为每个顶点指定编号，只需使用列表中的索引：

>>> list(enumerate(vs))
[(0, 'bag'), (1, 'cat'), (2, 'dog'), (3, 'fog'), (4, 'man'), (5, 'nut'), (6, 'rat')]
>>> i_by_v = {v: i for i, v in enumerate(vs)}
>>> i_by_v
{'bag': 0, 'cat': 1, 'dog': 2, 'fog': 3, 'man': 4, 'nut': 5, 'rat': 6}

让我们用它们的编号替换顶点：

>>> [(i_by_v[v1], i_by_v[v2]) for v1, v2 in edges]
[(4, 5), (4, 0), (5, 0), (6, 1), (2, 3), (2, 1), (2, 4), (3, 1), (3, 4), (1, 4)]

现在您可以使用任何您想要的图形算法。